Progress of statistics in applied animal behaviour science (1): General linear model

Abstract

In the research fields of applied animal behaviour science and livestock management, we sometimes meet the situation that normality of experimental data cannot be assumed. Since general linear model such as ANOVA or regression analysis assumes the normality, data transformation or non-parametric methods were conventionally applied for such data. However, the data transformation do not always ensure the normality of the transformed data, and statistical power of non-parametric methods become lower than those of parametric methods. As a possible alternative of these conventional methods, generalized linear model have become widely used. The generalized linear model is the extension of the general linear model and deal with various distribution including normal distribution. In this manuscript, we firstly introduce the concepts of general linear model and then show the examples of analysis with sample data and practical commands using free software “R”.